SIIS AI Password Strength Project (Starter)

This project is a starter implementation for your proposal: AI-driven password strength evaluation using machine learning and attack simulation.

What is included

1. Password feature extraction module.
2. Lightweight ML-style centroid classifier (dependency-light).
3. Brute-force, dictionary, and hybrid simulation module.
4. Flask web interface for password evaluation.
5. Training script with sample dataset.

Setup

cd c:\Users\Administrator\Documents\AYLF\siis_password_ai
python -m venv .venv
.venv\Scripts\Activate.ps1
pip install -r requirements.txt

Train model

python scripts\train_model.py

Run app

python app.py

Open http://127.0.0.1:5000.

How the app works (snippet)

User enters password
        |
        v
Feature Extraction
(length, digits, symbols, entropy, patterns)
        |
        v
ML Strength Prediction
(weak / medium / strong)
        |
        v
Attack Simulation
(brute force, dictionary, hybrid)
        |
        v
Risk Report
(risk level, crack-time estimate, improvement hints)

Presentation format (quick demo flow)

1. Problem: rule-based checkers can misclassify weak passwords.
2. Solution: AI + attack simulation framework.
3. Architecture: input -> feature extractor -> ML model -> simulator -> report.
4. Live Demo:
   • Test password123 (expected high risk).
   • Test C0mp!exKey2026 (expected lower risk).
5. Results: show predicted class, risk level, and crack-time bucket.
6. Impact: stronger user password habits and better policy enforcement.
7. Future Work: larger datasets, stronger ML models, real-time integration.

Suggested next improvements

1. Replace sample dataset with a larger labeled dataset.
2. Replace centroid classifier with Random Forest/XGBoost in full implementation.
3. Add confusion matrix and ROC curves to your report.
4. Add API endpoints for integration into registration systems.